We have read articles on the psychological characteristics of EQ-5D-5L, EORTC QLU-C10D, and general and cancer preference-based measurements, respectively. The author compares sensitivity to a known group, a common and independent approach to evaluating structural effectiveness. Here, five patient characteristics, including the family history of cancer, the time after treatment, the treatment method, the tumor stage, and the metastatic state, were selected as the “known group” verification. Nine national value sets were used for 10 EQ-5D-5L and QLU-C10D. The authors concluded that the EQ-5D-5L has a validity of the group known, and that it is better than QLU-C10D in discriminatory force among gastric cancer patients.
This conclusion is not valid based on the analysis and results presented. We believe it is important to emphasize some important issues, especially in the reported study, especially how to draw and the conclusions derived. We want to solve this problem politely as follows.
The author is the ability to detect the difference between the groups that are expected to differ in the discriminatory force of the measurement based on the conclusion of the known group hypothesis test. This requires a firm hypothesis that is guided by evidence or at least a certain patient group to show a difference in health utility and the theoretical basis for evidence or at least a well -known expert. In short, the hypothesis of a known group requires diligent justification to avoid misunderstandings for random or fake associations, that is, confusing variables. In addition, the observed results must be consistent with the hypothesis. Both conditions were violated here.
Regarding the interpretation of the test results, we must deal with two known group hypotheses. The first hypothesis assumed that the patient in the tumor stage would have a low health utility. The QLU-C10D results coincide with this hypothesis, comparing step I vs. II, III and IV. However, the EQ-5D-5L results showed the highest utility in the first stage and the highest advanced tumor stages. Therefore, these results are actually about EQ-5D-5L validity in this known group test. It is not a valid explanation that the author’s reasoning, the appropriate sample size can be responsible for the wrong direction in the EQ-5D-5L results. The appropriate sample size can be explained with sufficient certainty, but it is impossible to explain the effectiveness (statistically significant) effect in the opposite direction.
Similarly, the result of the second hypothesis was interpreted. In this paper, the author has a long time between the survey and the last treatment, so the health utility is lowered. However, the authors clarified in personal communication that typos have occurred here and have actually anticipated higher utility as time increased after treatment. The QLU-C10D showed a slight increase in both time intervals tested with all national value sets. However, the EQ-5D-5L results show an inconsistent pattern that does not clearly match the hypothesis of the expected direction of the effect. The first interval shows a significant increase in the utility, but the second interval often does not increase or decrease. These results were no longer discussed, but it was interpreted as a known group validation of EQ-5D-5L.
Regarding the justification of the hypothesis, we need to solve two additional “known” group variables. First, the author hypothesized that patients with cancer family history would have lower health utility than patients without patients. It is not clear why the family history of cancer affects the health dimensions such as mobility, pain, and depression in this specific direction. The authors did not provide theoretical basis for this hypothesis. Evidence of the literature on these effects can be found in a single paper using the same data set (1) as the paper discussed here. This fact disqualifies the use of the literature for justification in this study. Therefore, as long as there is no more evidence to prove the hypothesis, this is a prejudice approach.
We found problems such as lack of justification for the “known” difference and the author’s reason for the second hypothesis about the treatment of treatment. First, we knew it was difficult to understand what each group actually received. The author says “surgery” (N = 105), “chemotherapy” (chemotherapy ”(N = 34) and “both” (both ”(N = 104). However, the group is described as follows: “All participants were being treated, 209 participants had surgery (86.8%), 138, chemical radiation therapy (49.3%), and 34 were treated.” Second, we do not see a clear basis for hypotheses. The author has a higher utility of chemotherapy than surgery and the highest utility of patients receiving two therapies. In personal communication, the author explained that surgery is more aggressive with chemotherapy. However, considering the potential changes in each of each treatment, it is difficult to understand the reason why the author anticipates a particular difference and suggests them as “known” due to the combination of factors such as treatment characteristics (eg, chemotherapy, radiation, surgical degree), cancer phase or after treatment (2 and 3). Third, the authors responded personally that patients with treatment (surgery and chemotherapy) in China generally have better health. This accident line doubts the possibility of generalization of the results.
In summary, two of the five groups used (family history and treatment methods) were not considered a “known group” with sufficient evidence or theoretical basis. Each of the remaining three known groups (tumor stages, metastasis and time after treatment) was tested as a set of nine countries of EQ-5D-5L, respectively.N = 27), only 56% (N The test = 15) showed the results of the hypothesis. This is much lower than the threshold of the cosin guideline, where -here should follow the hypothesis to support the configuration effectiveness of more than 75% of the test (4 and 5). The QLU-C10D, evaluated using 10 value sets for each country, exhibited 29 (97%) of the 30 results that match the hypothesis. Therefore, the conclusion of this paper, the EQ-5D-5L, is not supported by the proposed data that the qlu-C10D surpasses the discriminatory force of gastric cancer patients and in this context. We think that the conclusion of this paper requires amendments to not mislead readers and potential users.
